Anterior versus Posterior Fusion Surgery for Cervical Ossification of the Posterior Longitudinal Ligament: A Systematic Review and Meta-Analysis

Abstract

Background

Cervical ossification of the posterior longitudinal ligament (OPLL) often requires surgical decompression with fusion. Both anterior decompression and fusion (ADF) and posterior decompression and fusion (PDF) are used; however, comparative evidence regarding their clinical outcomes remains limited and inconsistent. This study aimed to systematically review and synthesize evidence comparing ADF and PDF in patients with cervical OPLL.

Methods

A systematic search of Embase, the Cochrane Library, and MEDLINE was conducted from database inception to November 30, 2024. Randomized controlled trials and cohort studies directly comparing ADF and PDF were included. Outcomes of interest included neurological recovery based on the Japanese Orthopaedic Association score or recovery rate, complication rates, radiographic parameters, pain, reoperation, patient-reported outcome measures (PROMs), and healthcare costs. Pooled analyses were performed using weighted mean differences or odds ratios with 95% confidence intervals. Certainty of evidence was assessed using the Grading of Recommendations Assessment, Development and Evaluation framework.

Results

Eleven studies met the inclusion criteria. The overall certainty of evidence was rated as very low. Pooled point estimates favored ADF for neurological recovery compared with PDF, although differences did not reach statistical significance. Similar findings were observed in subgroup analyses of patients with K-line (–) status or high canal-occupying ratios (≥50% to 60%). Perioperative complications were more frequent after ADF, whereas neurological complications occurred more often after PDF; however, neither comparison showed statistically significant differences. No significant differences were identified in pain reduction, radiographic outcomes, reoperation rates, or PROMs. One study reported higher healthcare costs in the PDF group. Substantial heterogeneity was present across most analyses.

Conclusions

Current evidence does not demonstrate a clear advantage of either anterior or posterior fusion strategies for cervical OPLL. Given the substantial heterogeneity and very low certainty of evidence, differences between approaches should be interpreted with caution. Further well-designed comparative studies are required to determine the optimal fusion strategy.

Introduction

Cervical ossification of the posterior longitudinal ligament (OPLL), which is more prevalent in East Asian populations, is a progressive disorder that compresses the spinal cord¹⁾. With disease progression, patients may develop motor and sensory deficits in the trunk and extremities, as well as autonomic dysfunction such as bladder and bowel disturbances. In many cases, surgical decompression is required to prevent further neurological deterioration, and favorable outcomes have generally been reported^2,3). Surgical strategies can be broadly categorized into decompression alone and decompression with fusion. Laminoplasty, a representative decompression procedure without fusion, is technically straightforward and widely performed; however, it may be associated with postoperative axial neck pain and progressive ossification over time, leading to the frequent selection of fusion procedures^4,5). In particular, patients with a negative K-line or a high canal-occupying ratio often experience insufficient neurological recovery after laminoplasty alone, and fusion surgery should be considered in such cases^6-8).

Fusion surgery can be performed via either an anterior or a posterior approach. Nevertheless, most available reports are limited to small sample sizes, single-center studies, or retrospective designs. Consequently, reported outcomes are inconsistent, and the overall strength of evidence from individual studies remains insufficient. Therefore, this study aimed to conduct a systematic review and meta-analysis of studies comparing anterior and posterior decompression and fusion (PDF) procedures for cervical OPLL. By integrating data across studies, this approach can increase statistical power, provide more precise estimates of treatment effects and complication rates, and clarify the current state of evidence.

Materials and Methods

Search strategies

An extensive literature search was conducted in the Cochrane Library, MEDLINE, and Embase using Medical Subject Headings (MeSH) complemented with a free-text search. The following search terms were used: (1) “cervical spine” AND “ossification of the posterior longitudinal ligament”; (2) “fusion surgery”; (3) “posterior approach”; (4) “posterior fusion”; (5) “anterior approach”; and (6) “anterior fusion.” The final search query was constructed as (1) AND (2) AND [(3) OR (4)] AND [(5) OR (6)]. The detailed search strategy is provided in the Supplemental data.

The aim of this search was to identify randomized controlled trials (RCTs) and observational studies comparing anterior decompression and fusion (ADF) with PDF for cervical OPLL. Reference lists from previous reviews were also checked to identify additional articles that met the inclusion criteria. The language was restricted to English, and the publication period extended from the inception of each database to November 30, 2024.

The following inclusion criteria were adopted: (1) RCTs, controlled clinical trials, or cohort studies; (2) study population: cervical OPLL; (3) intervention: ADF and PDF; (4) surgical outcomes: neurological outcomes, patient-reported outcome measures (PROMs), pain outcomes, radiological outcomes, economic outcomes, and complications; and (5) available comparative data for ADF and PDF. Exclusion criteria were as follows: studies including (1) patients with trauma, tumor, or infection; (2) patients with previous cervical surgery; (3) patients undergoing a combined anterior and posterior surgical approach; and (4) case reports, non-human studies, non-comparative studies, and review articles.

Ethics approval was not required for this study.

Data extraction and quality assessment

We extracted the following outcome measures from the included studies: neurological function, assessed using the Japanese Orthopaedic Association (JOA) scoring system for cervical myelopathy⁹⁾ and the neurological recovery rate, calculated according to Hirabayashi's method¹⁰⁾; PROMs related to quality of life, including the 36-Item Short Form Health Survey (SF-36), the JOA Cervical Myelopathy Evaluation Questionnaire (JOACMEQ), the Neck Disability Index (NDI), and the Oswestry Disability Index (ODI); neck pain assessed using the Visual Analogue Scale (VAS); radiographic parameters, including the C2-C7 angle and cervical range of motion (ROM); outcomes related to healthcare costs and resource utilization; and surgical outcomes, such as complication and reoperation rates. The level of evidence was assessed according to the Grading of Recommendations Assessment, Development and Evaluation (GRADE) scale¹¹⁾.

Data analysis

For the pooled effects, the weighted mean difference was calculated for continuous variables according to the consistency of measurement units, and the odds ratio (OR) was calculated for dichotomous variables. Continuous variables are presented as mean differences and 95% confidence intervals (CIs), whereas dichotomous variables are presented as ORs and 95% CIs. For data presented as the median and range, the values of means and standard deviations were estimated using Hozo's method¹²⁾.

Regarding the evaluation of surgical outcomes in patients with more severe OPLL, cases with K-line (–)⁷⁾ and those with a canal-occupying ratio defined as ≥50% or ≥60% in the original studies were combined in the subgroup analysis. These variables are generally recognized as indicators of advanced disease, and separating them would have markedly reduced the number of eligible studies, thereby precluding a meaningful pooled analysis.

All statistical tests were performed using SPSS software (Version 30.0; IBM Corporation, Armonk, NY, USA) and Review Manager (Version 5.4; The Cochrane Collaboration, London, UK).

Results

Search results

Eleven studies were obtained from Embase, the Cochrane Library, MEDLINE, and other sources^8,13-22). A flowchart of the study selection process is shown in Figure 1. There were one RCT, one prospective non-randomized study, and nine retrospective cohort studies (Table 1). Quality assessment of the included studies was performed. According to the GRADE scale, the quality of all included studies was rated as very low (Table 2).

Figure 1.

Flow chart of study selection.

Flow diagram of study identification, screening, eligibility assessment, and inclusion. A total of 529 records were identified from the Cochrane Library (n=14), MEDLINE (n=271), and Embase (n=244), with one additional record identified from other sources. After screening, 18 full-text articles were assessed for eligibility, 11 studies were included in the qualitative synthesis, and 10 studies were included in the quantitative synthesis (meta-analysis).

Table 1.

Baseline Characteristics of the Studies.

Author	Year	Design	Approach	N	Gender (M, F)	Age (y)	Surgical methods	Levels	Follow-up (months)
Chen et al.15)	2012	Retrospective	Ant	91	63, 28	48.7±1.4	ACCF	2.7±0.2	48 (at least)
			Post	32	19, 13	52.6±1.7	LAMF	4.2±0.1	48 (at least)
Lin et al.17)	2012	Retrospective	Ant	26	15, 11	54.7±13.2	ACCF	3.92±0.50	36.3±6.4
			Post	30	17, 13	56.2±14.1	LAMF	4.27±0.75	37.6±6.7
Koda et al.8)	2016	Retrospective	Ant	15	10, 5	57.7±5.0	ACCF	N/A	58.6±30.3
			Post	17	14, 3	65.0±11.8	LAMF	N/A	42.0±22.8
Yoshii et al.22)	2016	Retrospective	Ant	39	31, 8	61.1±8.5	ACCF	3.1±0.9	44.5±18.8
			Post	22	18, 4	60.6±12.8	LAMF	5.3±2.0	37.2±16.3
Yang et al.20)	2019	Retrospective	Ant	14	N/A	N/A	ACCF	N/A	N/A
			Post	47	N/A	N/A	LAMF	N/A	N/A
Yoshii et al.21)	2020	Retrospective	Ant	854	602, 252	62.6±11.2	ACCF/ACDF	N/A	N/A
			Post	854	602, 252	62.4±10.8	LAMF	N/A	N/A
Park et al.19)	2021	Retrospective	Ant	18	10, 8	55.5±8.3	ACCF/ACDF	N/A	12.67±18.76
			Post	10	8, 2	54.5±6.0	LAMF	N/A	14.23±23.68
Nagoshi et al.18)	2023	Prospective	Ant	27	15, 12	60.8±10.7	ACCF/ACDF	3.4±1.1	24
			Post	27	20, 7	59.4±8.9	LAMF	4.9±1.2	24
Lee et al.16)	2023	Retrospective	Ant	387	228, 159	57.7±10.1	ACCF/ACDF	N/A	N/A
			Post	320	170, 150	58.5±10.6	LAMF	N/A	N/A
Cao et al.13)	2023	RCT	Ant	22	11, 11	52.91±7.41	ACCF	N/A	24
			Post	18	15, 3	57.17±7.97	LAMF	N/A	24
Chen et al.14)	2024	Retrospective	Ant	24	18, 6	52.88±8.79	ACCF/ACDF	N/A	148.8±17.6
			Post	24	19, 5	56.25±9.44	LAMF	N/A	141.8±16.2

ACCF: anterior cervical corpectomy and fusion; ACDF: anterior cervical discectomy and fusion; Ant: anterior; LAMF: laminoplasty/laminectomy with fusion; N/A: not available; Post: posterior; RCT: randomized controlled trial

Table 2.

Quality Assessment.

	Study design	Participants (studies)	Risk of bias	Inconsistency	Imprecision	Indirectness	Publication bias	Other consideration	Quality of evidence
Neurological improvement	RCT	40 (1)	Serious (-1)	N/A	Serious (-1)	Not serious	Not serious	None	Very low
	OS	463 (8)	Serious (-1)	Very serious (-2)	Not serious	Not serious	Not serious	None	Very low
Neurological improvement (severe cases)	RCT	10 (1)	Serious (-1)	N/A	Very serious (-2)	Not serious	Not serious	None	Very low
	OS	237 (5)	Serious (-1)	Very serious (-2)	Serious (-1)	Serious (-1)	Not serious	None	Very low
Neck pain	RCT	40 (1)	Serious (-1)	N/A	Serious (-1)	Not serious	Not serious	None	Very low
	OS	191 (4)	Serious (-1)	Not serious	Serious (-1)	Not serious	Not serious	None	Very low
Overall complications	RCT	40 (1)	Serious (-1)	N/A	Serious (-1)	Not serious	Not serious	None	Very low
	OS	2590 (6)	Serious (-1)	Serious (-1)	Not serious	Not serious	Not serious	None	Very low
Neurological complications	RCT	40 (1)	Serious (-1)	N/A	Serious (-1)	Not serious	Not serious	None	Very low
	OS	2590 (6)	Serious (-1)	Serious (-1)	Serious (-1)	Not serious	Not serious	None	Very low
Reoperation	RCT	40 (1)	Serious (-1)	N/A	Serious (-1)	Not serious	Not serious	None	Very low
	OS	2335 (2)	Serious (-1)	Not serious	Not serious	Not serious	Not serious	None	Very low
Alignment	RCT	40 (1)	Serious (-1)	N/A	Serious (-1)	Not serious	Not serious	None	Very low
	OS	175 (4)	Serious (-1)	Very serious (-2)	Serious (-1)	Not serious	Not serious	None	Very low
ROM	RCT	40 (1)	Serious (-1)	N/A	Serious (-1)	Not serious	Not serious	None	Very low
	OS	110 (2)	Not serious	Not serious	Very serious (-2)	Not serious	Not serious	None	Very low

N/A: not available; OS: observational study; RCT: randomized controlled trial; ROM: range of motion

Evaluation of the RCT

Cao et al.¹³⁾ reported the only available RCT. In their study, ADF (n=22) demonstrated a significantly greater improvement in the JOA recovery rate compared with PDF (n=18) (68.10±34.23 vs. 38.39±44.18; p=0.016). Among patients with K-line (–), the anterior approach (n=6) also yielded significantly superior recovery rates compared with the posterior approach (n=4) (82.67±16.73 vs. 37.50±26.30; p=0.016). No significant differences were observed between the two approaches in terms of complication rates, reoperation rates, reduction of neck pain, quality of life as assessed by the NDI, or changes in cervical alignment and ROM.

Regarding the level of evidence, because these findings are derived from a single RCT, the certainty of evidence should be initially rated as low. In addition, the study did not describe whether outcome assessors were blinded during postoperative follow-up, and the sample size was relatively small. For these reasons, the strength of the evidence was judged to require further downgrading.

Evaluation of observational studies

A meta-analysis was conducted using 10 observational studies, excluding the single RCT. In the pooled analysis, the mean age was 60.0±11.1 years in the ADF group (n=1,481) and 60.9±10.9 years in the PDF group (n=1,336), with the difference reaching statistical significance (p=0.03). The proportion of male patients was comparable between the two groups (ADF: 66.4% vs. PDF: 64.1%; p=0.74). The number of fused levels was significantly lower in the ADF group (3.1±0.8) compared with the PDF group (4.6±1.2; p<0.01). The mean follow-up period was similar between groups (ADF: 60.3±49.3 months vs. PDF: 60.3±48.4 months; p=0.99).

Recovery rate of JOA score

Eight observational studies were included in the pooled analysis of postoperative neurological recovery based on the JOA recovery rate (Figure 2A). The pooled standardized mean difference (SMD) for neurological recovery favored ADF compared with PDF (ADF, n=254; PDF, n=209; SMD, −0.93; 95% CI, −1.86 to 0.01); however, the difference did not reach statistical significance (p=0.05). Substantial heterogeneity was observed across studies (I²=95%).

Figure 2.

Recovery rate of JOA score.

JOA: Japanese Orthopaedic Association

In patients with K-line (–) or with a canal-occupying ratio defined as ≥50% or ≥60%, five studies were included in the subgroup analysis (Figure 2B). In this subgroup, the pooled point estimates again favored the ADF group compared with the PDF group (ADF, n=129; PDF, n=108; SMD, −1.24; 95% CI, −2.51 to 0.02), although the CIs were wide and the difference did not reach statistical significance (p=0.05). Considerable heterogeneity was observed across studies (I²=94%).

Change in cervical VAS

Four observational studies were included in the pooled analysis of pre- to postoperative changes in cervical VAS scores (Figure 3). No significant difference was observed between the ADF group (n=83) and the PDF group (n=108), with a pooled SMD of 0.04 (95% CI, –0.33 to 0.41; p=0.84). Low heterogeneity was detected across studies (I²=31%).

Figure 3.

Difference in cervical VAS.

VAS: visual analogue scale

Complication rates

Six observational studies were included in the pooled analysis of overall complications (Figure 4A). The complication rate was higher in the ADF group (190/1320) than in the PDF group (107/1270). The pooled OR was 1.67 (95% CI, 1.00–2.77; p=0.05), which did not reach statistical significance. Moderate heterogeneity was observed across studies (I²=54%).

Figure 4.

Complication rates.

In the analysis of neurological complications, no significant difference was found between ADF (17/1320) and PDF (24/1270), with a pooled OR of 0.50 (95% CI, 0.11–2.29; p=0.37; I²=69%) (Figure 4B).

Reoperation rates

Two observational studies were included in the pooled analysis of reoperation rates (Figure 5). Reoperation occurred in 30 of 1,161 patients in the ADF group and 29 of 1,174 patients in the PDF group. The pooled OR was 1.04 (95% CI, 0.62–1.76; p=0.87), with no evidence of heterogeneity (I²=0%).

Figure 5.

Reoperation rates.

Evaluation of cervical spine radiographs

Four observational studies were included in the pooled analysis of differences in cervical alignment (C2-C7 Cobb angle) between pre- and postoperative measurements (Figure 6A). No significant difference was found between ADF (n=99) and PDF (n=76), with a pooled SMD of –0.60 (95% CI, –1.50 to 0.31; p=0.20). Substantial heterogeneity was observed across studies (I²=87%).

Figure 6.

Evaluation of cervical spine radiographs.

Two studies assessed differences in cervical ROM (Figure 6B). The pooled analysis showed no significant difference between ADF (n=53) and PDF (n=57), with a pooled SMD of –0.32 (95% CI, –0.70 to 0.05; p=0.09), and no heterogeneity was detected (I²=0%).

PROMs

One study utilized the JOACMEQ¹⁸⁾, and another employed the ODI and SF-36 for evaluation¹⁹⁾. As these instruments were not consistent across studies, a meta-analysis could not be performed. In both reports, no significant differences in postoperative outcomes were observed between the ADF and PDF groups.

Healthcare costs

Only one study evaluated healthcare costs, reporting that among 854 matched cases in each group, the mean cost was $22,768 in the ADF group and $30,663 in the PDF group (p<0.001)²¹⁾. No other studies assessed this outcome, and therefore, a meta-analysis could not be performed.

Discussion

Several systematic reviews have previously compared the clinical outcomes of anterior versus posterior approaches for cervical OPLL. However, most of these analyses have included posterior procedures without fusion, such as laminoplasty or laminectomy^23-26). When considering surgical strategies, one important perspective is whether fixation is necessary or whether decompression alone without fusion is sufficient, a decision often guided by factors such as the K-line, occupying ratio of ossification, and cervical alignment^7,8,22). Fusion procedures aim to achieve bone union, whereas decompression procedures such as laminoplasty are designed to preserve motion, and thus the fundamental surgical objectives differ. Therefore, in the present study, we focused exclusively on fusion surgeries and investigated the differences in surgical outcomes between anterior and posterior approaches.

In the quality assessment of the included studies, multiple domains were rated as serious (Table 2). In many reports, the sample size of either the ADF or PDF group was limited to approximately 20-30 patients (Table 1), and adjustments for potential confounding factors were insufficient. Furthermore, postoperative rehabilitation and pharmacological management were often not clearly described, and the follow-up periods were not standardized, raising concerns regarding their potential impact on clinical outcomes. Only one randomized controlled trial was identified, while the remaining 10 studies were observational in nature. Considering these multiple serious limitations, the overall certainty of evidence for many of the outcomes must be judged as very low.

Based on this meta-analysis, neurological recovery following ADF was greater than that following PDF; however, the pooled effects did not reach statistical significance (Figure 2A and B). Although the point estimates consistently favored the anterior approach—particularly in patients with K-line (–) status or those with a high canal-occupying ratio—the wide CIs and substantial heterogeneity across studies limit the strength of these findings. Therefore, though anterior decompression may provide theoretical advantages due to the anatomical location of the ossified ligament^8,14,15), the present evidence is insufficient to draw firm conclusions regarding its superiority.

Similarly, the higher incidence of perioperative complications observed in the ADF group (Figure 4A) demonstrated only borderline significance, and the certainty of this estimate is low. Dysphagia, airway-related issues, cerebrospinal fluid leakage, and graft-related problems remain well-recognized complications associated with the anterior approach^{8,14-16,18,19,21,22,27,28)}; however, the lack of statistical significance in the pooled analysis indicates that this difference should be interpreted cautiously. Neurological complications, including C5 palsy, were numerically more frequent after PDF than after ADF, although this difference did not reach statistical significance (Figure 4B).

Postoperative pain reduction, changes in cervical alignment, reoperation rates, and PROMs were evaluated in only a limited number of studies, making it difficult to draw firm conclusions; however, no significant differences were observed between the two groups (Figures 35 and 6A). With respect to healthcare costs, only one study addressed this outcome, reporting significantly higher costs in the PDF group compared with the ADF group²¹⁾. Nevertheless, as this finding derives from a single report conducted within the framework of a specific regional healthcare system, its generalizability is limited, and further research is warranted to establish more universal conclusions.

In the meta-analysis of observational studies, a statistically significant difference in patient age was observed between groups. However, the mean age was 60.0 years in the ADF group and 60.9 years in the PDF group, with an absolute difference of only 0.9 years, which is not considered clinically meaningful. In contrast, a notable difference was observed in the extent of fusion, with shorter fusion segments in the anterior approach compared with the posterior approach. This finding is consistent with technical considerations, as posterior fixation is generally more feasible for the upper cervical spine, and previous reports have similarly indicated that PDF tends to involve a greater number of levels^15,18,22,28). In line with these observations, our results suggested lower postoperative cervical ROM in the PDF group, although this finding was based on a limited number of studies and did not reach statistical significance (Figure 6B). This finding may warrant consideration during surgical decision-making and should be shared with patients as part of the informed consent process.

Taken together, the present systematic review demonstrates that the majority of available comparative studies on anterior versus posterior fusion for cervical OPLL are characterized by very low certainty of evidence according to the GRADE framework. Across key outcomes, including neurological recovery and complication rates, no statistically significant differences were consistently observed between approaches, and substantial heterogeneity was present. Moreover, PROMs, radiographic outcomes, and healthcare costs have been insufficiently investigated in the existing literature. These limitations preclude the ability to draw definitive conclusions regarding the superiority of either surgical approach. Importantly, the absence of statistically significant differences should not be interpreted as evidence of equivalence between anterior and posterior fusion, but rather as a reflection of the current limitations and imprecision of the evidence base. In this context, surgical decision-making must rely on careful consideration of patient-specific and disease-specific factors, including OPLL morphology, severity, alignment, and perioperative risk profiles, rather than on an assumption of overall superiority of one approach. Although further accumulation of high-quality, multidimensional evidence is clearly required, the present findings underscore the need for cautious and individualized selection of surgical strategies for cervical OPLL within the constraints of current clinical practice.

In conclusion, the pooled point estimates favored ADF for neurological recovery and PDF for perioperative safety, whereas neither comparison reached statistical significance. Nonetheless, the overall evidence remains limited by heterogeneity and study quality, underscoring the need for more rigorous comparative research.

Author Contributions: N.N. was responsible for literature selection, data analysis, and manuscript drafting. T.H., H.T., K.M., K.W., S.M., Y.O., G.Y., S.I., M.T., H.C., M.K., M.I., Y.K., and M.Ya. critically reviewed the study content. M.Yo. provided guidance on the methodological approach. TY supervised the overall conduct of the study.

Conflicts of Interest: The authors declare that there are no relevant conflicts of interest.

Ethical Approval: This study is a systematic review and meta-analysis based exclusively on previously published studies. As it does not involve the collection or analysis of any individual-level patient data, institutional review board (IRB) approval was not required.

Informed Consent: As this study did not involve direct interaction with patients or the use of identifiable personal information, informed consent was not applicable.

Disclaimer: Yoshiharu Kawaguchi is the Editor-in-Chief of Spine Surgery and Related Research and a member of the journal's Editorial Committee. He was not involved in the editorial evaluation or decision to accept this article for publication.

Funding Statement

This work was supported by the Health and Labour Sciences Research Grant from the Ministry of Health, Labour and Welfare, Japan (23FC1046), and by the Japan Agency for Medical Research and Development (AMED) (23ek0109541h0003).